Testing the Neocon Agenda: Democracy in Resource-Rich Societies
Paul Collier and Anke Hoeffler,
Department of Economics,
University of Oxford
November, 2007
Abstract:
Resource-rich countries have tended to be autocratic and also have tended to use their
resource wealth badly. The neoconservative agenda of promoting democratization in
resource-rich countries thus offers the hopeful prospect of a better use of their
economic opportunities. This paper examines whether the effect of democracy on
economic performance is distinctive in resource-rich societies. We show that a priori
the sign of the effect is ambiguous: resource rents could either enhance or undermine
the economic consequences of democracy. We therefore investigate the issue
empirically. We first build a new data set on country-specific resource rents, annually
for the period 1970-2001. Using a global panel data set we find that in developing
countries the combination of high natural resource rents and open democratic systems
has been growth-reducing. Checks and balances offset this adverse effect. Thus,
resource-rich economies need a distinctive form of democracy with particularly strong
checks and balances. Unfortunately this is rare: checks and balances are public goods
and so are liable to be undersupplied in new democracies. Over time they are eroded
by resource rents.
This document is an output from a project funded by the UK Department
for International Development (DFID) for the benefit of developing
countries. The views expressed are not necessarily those of DFID.
JEL Classification: O11, O40, Q38
Keywords: political economy, natural resources, growth
1. Introduction
Many resource-rich countries have suffered from the ‘resource curse’: the failure to
harness commodity revenues for sustained growth. In this paper we investigate
analytically and empirically whether this curse is ameliorated by democracy. The
question is at the intersection of two large and active literatures, one accounting for
the resource curse and the other investigating the economic consequences of
democracy. On the intersection, however, there is to date very little. Yet the
intersection is of considerable practical concern. The recent rise in commodity prices
has generated revenue booms in commodity exporters that were last experienced in
the 1970s. The widespread failure to harness those booms for sustained growth is the
empirical basis for the resource curse: hence, whether the present booms will repeat
history is of first order importance for a large group of low-income societies. Between
the two resource booms one striking institutional change has occurred: resource-rich
countries are now on average more democratic.i The importance of institutions and
the scope for changing them are now central controversies in development economics.
Thus, our investigation of the economic consequences of democratization in resourcerich countries nests within this larger debate.
The ‘neocon agenda’ of the United States, combined with two other influences, has
increased the prevalence of resource-rich democracies. The agenda diagnosed the
perceived ills of the Middle East as being due to its lack of democracy. Thus, Selden
(2004) defines the neocon agenda as using:
‘American power to reshape the global environment in the name of a set of
liberal democratic ideals. It is their belief that this will make the United States
more secure by reducing the seemingly intractable problems of the Middle East,
thus getting at some of the root causes of terrorism.’
2
Democracy as promoted by the American government partly by hard power, as in
Iraq, and partly by soft power, exemplified by the first multi-party Egyptian elections
held in 2005. The lack of democracy in the Middle East is itself related to the
resource abundance of the region: Ross (2001) shows that resource-rich countries are
systematically less democratic.
In addition to the neocon agenda the ‘third wave’ of
democratization following the fall of the USSR spread to some resource-rich
countries. Additionally, some of the resource discoveries triggered by high prices
have been in countries that had already democratized.
The concept of the resource curse was initially triggered by particular graphic
instances such as Nigeria. There is now a substantial literature exploring both the
evidence for the curse and the mechanisms by which it is generated (for example
Sachs and Warner (2005), Auty (2001), Gylfason (2001)). There is overwhelming
statistical evidence that countries rich in natural resources experienced lower growth
rates since 1960. Although the initial statistical evidence was based on cross-section,
it is now supported by time series analysis of panel data (Collier and Goderis, 2007,
2007a). The explanation for the curse has also evolved. The initial explanation was
the purely economic process of Dutch disease (Corden and Neary, 1982) whereas
more recent explanations have been framed in terms of political economy. Tornell and
Lane (1999) proposed a ‘voracity effect’ whereby the high value of resource rents
would induce competing political groups to ‘gauge’ in the process not merely wasting
the rents as in conventional rent-seeking, but actually reducing income. The
celebrated work of Acemoglu, Johnson and Robinson (2000) also featured a key
adverse role for resource rents: inducing institutions suited to extracting rents rather
than to the provision of public goods. They proposed that these types of institutions
were both distinctive and highly persistent, generating divergent paths of economic
development. The two articles to which our work is most closely related are
3
Robinson, Torvik and Verdier (2006) and Mehlum, Moene and Torvik (2006) since
they address not only why political processes might be dysfunctional in the context of
resource rents, but suggest that this may be remedied through appropriate institutions.
Robinson, Torvik and Verdier develop a theory of patronage politics in the context of
resource wealth and suggest that this dysfunctional behavior may be restrained by
good institutions. Mehlum, Moene and Torvik find some empirical support for the
idea that institutions are particularly important in the context of natural resources but
do not investigate specifically which institutions are important. Both articles rely upon
the Sachs and Warner cross-section data set. The present paper extends these studies
both analytically, by proposing a more detailed mechanism of political failure, and
empirically by using a panel data set and unbundling the concept of ‘institutions’.
While the resource curse literature has thus evolved to the stage where institutional
weaknesses are regarded as central to the explanation of the curse, it has not yet
specifically related these weaknesses either to whether the polity is autocratic rather
than democratic, or to variations in democratic design. Yet, as we have noted,
democratization has been the main recent institutional innovation in resource-rich
countries and in other contexts its effects on economic performance, including the
effects of variations in its design, and the possibility of reverse causality from
development to democracy, have been intensively studied (Lipset, 1959, Przeworski
and Limongi, 1993), Acemoglu, Johnson, Robinson, and Yared, 2008). Overall, the
net effect of democracy on economic development is far from clear-cut. Drazen
(2000) concludes from a survey of the literature that there is no clear effect.
Acemoglu et al. (2008) suggest that different countries are on different long term
trajectories, some of which lead to both development and democracy, whereas others
4
lead to development without democracy. They do not find any evidence of a causal
link between income and democracy but provide some statistical evidence to show
that omitted variables, most likely historical variables, jointy determine the political
as well as economic development paths. Barro (1996) finds evidence for a non-linear
effect: democracy is beneficial at ‘intermediate levels’.
None of this literature considers resource rents. For example, it is entirely omitted in
the comprehensive survey of theory and evidence by Feng (2003). Yet this neglect is
surprising. One of the few clear results of the literature on the economic effects of
democracy has been to highlight the problem posed by the shortening of government
horizons introduced by the elections for those public decisions that require a longterm view. For example, Tavares and Warziarg (2001) find that democracies have
systematically lower investment than autocracies. High investment is central to
economic management in resource-rich countries since extraction depletes assets
which must be replaced if increased consumption is to be sustained. While this
suggests that democracy might be problematic for resource-rich countries, this
adverse effect might be overwhelmed by both the enhanced accountability of public
revenues, and by the widening of the power base that democracy facilitates. Or the
two opposing effects might broadly neutralize each other, leaving no clear
relationship between democracy and economic performance in resource-rich
countries.
In Section 2 we consider the mechanisms by which resource rents might either
enhance or undermine the contribution of democracy to economic development. We
show that a priori either is possible so that the issue can only be resolved empirically.
5
In Section 3 we explain a new empirical measure of the rents from natural resource
exports, country-by-country, for the period 1970-2001. This is a substantial advance
on the empirical proxy that has been standard in the literature, namely the value of
primary commodity exports. Evidently, rents differ radically both between different
commodities and are not proportional to changes in prices. In Section 4 we use this
measure to investigate whether the effect of democracy upon growth is altered by the
presence of natural resource rents. We find a large adverse interaction of natural
resource rents and electoral competition and a large positive interaction of natural
resource rents and checks and balances. We then investigate the routes by which
electoral competition and checks and balances might have these effects, and finally
show that over the long term checks and balances are endogenous to resource rents.
Section 5 concludes.
2. How might resource rents affect the economic consequences of democracy?
Potentially democracy might be either more or less advantageous for economic
performance if a society has large natural resources. While in the long run democracy
is itself liable to be endogenous to resource rents, Smith (2004) plausibly suggests that
because institutions usually pre-exist resource discoveries, the effects of the rents are
likely to be dependent upon this prior institutional variation. We consider some
possible mechanisms that would work in each direction.
Mechanisms that enhance the benefits of democracy
6
We suggest two mechanisms by which democracy might lead to a differential
improvement in economic performance. Both the economic consequences of
autocracy may be made absolutely worse by resource rents, and the economic
consequences of democracy may be made absolutely greater.
Autocrats may be particularly predatory in the presence of resource rents. The
reasoning is analogous to the famous distinction made by Olson between the roving
and the stationary bandit (Olson, 1993). Whereas the roving bandit snatches whatever
he can without regard to future consequences for the economy, the stationary bandit
must limit predation to a rate that is sustainable. The same argument applies if bandit
actions can extend to investments that expand the economy: the roving bandit does
not invest, but the stationary bandit invests. Being atypically immobile and highly
taxable, natural resource rents reduce the need for an autocrat to invest in the growth
of the non-resource private economy. This is indeed close to the explanation of
Acemoglu, Johnson and Robinson (2000) as to why autocratic colonial governments
were made dysfunctional by the presence of resource rents. An autocrat in resourcescarce economies, whether the British government in colonial Kenya or the
communist government of modern China, must develop the private economy if it is to
generate significant tax revenues and for this it must provide public goods. In
contrast, an autocrat in resource-rich central Africa, whether King Leopold or
President Mobutu, can generate revenues without the provision of public goods.
Conversely, democratic governments may be particularly useful in societies with
large resource rents. With the sole exception of America, such rents accrue in large
part to the government so that resource-rich states have an atypically high level of
7
public spending. Suppose, not implausibly, that democracy enhances the
accountability of public spending to citizens, preventing it from being captured by
powerful minorities. In this case, the pay-off to democracy will be approximately
proportional to the share of public spending in GDP and so differentially greater in
resource-rich societies. Accountable public spending would be far more beneficial to
Angola, where resource revenues alone are a more than half GDP, than to Uganda
where public spending in total is only a fifth of GDP.
While we do not develop further these sketches of why democracy might be
particularly valuable for resource-rich countries, they do not seem to be readily
dismissible as possibilities.
Mechanisms that undermine the benefits of democracy
We now develop at greater length an exposition of a mechanism that works in the
opposite direction. We devote greater length to it partly because it is less obvious than
the above mechanisms and also because, based on the results of Section 4, it appears
more likely to be correct.
We focus on the functioning of democracy and consider how politicians use public
resources subject to restraints. Potentially, these resources can be used either for the
provision of public goods or for private patronage. In a well-functioning polity
politicians who divert public resources into private patronage suffer both electoral
defeat and prosecution. We develop a simple model of these restraints, showing how
they may be undermined by natural resource rents.
8
In the absence of effective legal restraint political parties face a choice of technologies
in how they attract votes. They can adopt the conventional mode of presenting
programs to voters which commit to the provision of public goods. Alternatively, they
can directly buy votes. In mature democracies vote-buying is not a viable electoral
strategy for four reasons. Contracts with voters would be non-enforceable due to
secret ballots. Bribe transactions would be liable to criminal prosecution. They would
antagonize other voters. Finally, they would be too expensive to be affordable by
political parties. However, in the conditions typical of developing countries these four
inhibitions may not apply. Bribe contracts may indeed be enforceable because they fit
naturally into a prevailing culture of reciprocal exchange: bribes are regarded as gifts
which rightly attract obligations (Githongo, 2006). The police and courts may not be
sufficiently independent of the political process to prosecute vote-buying. In a
prevailing atmosphere of political corruption even electors may find vote-buying
acceptable: it is the only benefit they can realistically expect from participation in the
democratic process. Finally, political parties may have access to very large sources of
finance.
Vote-buying is indeed common in developing countries so political parties evidently
regard it as cost-effective. Vicente (2007) conducts a unique randomized analysis of
vote-buying in the resource-rich country of Sao Tome, Principe, and finds that it is
both widespread and effective in inducing voters to change allegiance. Another
widespread strategy that is analytically similar to vote-buying is for the politician to
hire a militia which then targets opposition supporters, discouraging them from
voting. Since the act of voting is readily observable and identity politics increases the
9
observability of allegiance, targeted intimidation may also be a cost-effective use of
political finance.
Besley (2006) shows analytically that the weaker is voter information about
government performance, and the more that voting is pre-determined by identity, the
less traction is available to an honest politician whose intentions are congruent with
voters. In these conditions corrupt politicians with dissonant intentions may find votebuying and intimidation highly effective. They can be targeted to the minority of
swing voters. Indeed, if many voters follow the instructions of community leaders, by
bribing them politicians can purchase votes ‘wholesale’. A core argument of Besley is
that the motivation of politicians is endogenous to these underlying electoral
conditions: where honest politicians with congruent intentions stand little chance of
election they will not come forward as candidates. We imagine a political contest in
which the underlying conditions completely discourage honest politicians: all the
candidates have dissonant intentions and are prepared to be dishonest. Expenditure on
bribery and intimidation is cost-effective in election campaigns and will predominate
to the extent permitted by political finance. This in turn depends upon how much
public revenue can be embezzled from its proper uses. To the extent that public
spending cannot be embezzled, even corrupt politicians will find themselves presiding
over systems which deliver public goods since these provide some electoral
advantage, despite rather than because of their intentions. In effect, public revenues
have two different degrees of electoral potency: embezzled money is high-powered
because it can be used for bribes, while money that cannot be embezzled can only be
used for the less cost-effective strategy of providing public goods. Thus, given our
assumptions, public goods are provided not because politicians need to do so in order
10
to win votes, but because the checks and balances present in the system prevent them
from diverting all revenues to patronage.
Hence, the key issue is the determination of the checks and balances that limit the
embezzlement of public resources. Effective checks and balances are themselves a
type of public good: to be effective, restraints need to be implemented through a
continuous process of public scrutiny. To endogenize this process we introduce a
relationship in which citizens are provoked into scrutiny by taxation. This relationship
is exemplified in the central demand of the American revolution: ‘no taxation without
representation’. As a proposition in political science it is most closely identified with
Tilly (1975), being central to his celebrated explanation of the emergence of
government accountability to citizens in Europe. More recently the proposition has
been central to the work of Moore (see, for example, Moore, 1998). Brautigam and
Knack (2004) provide a clear recent application in their discussion of the problems
associated with aid: ‘when revenues do not depend on the taxes raised from citizens
and businesses, there is less incentive for government to be accountable to them’
(p.265). Resource rents are a non-tax revenue somewhat analogous to aid. Ross
(2004) investigates empirically the link between taxation and representation.
Consistent with the Tilly hypothesis he finds that the larger is the share of government
expenditure which is financed through taxation the more likely is the government to
become representative.
We now develop a simple model that incorporates the above behavioral relationships.
Politicians would like to tax heavily in order to generate revenue for patronage, but
they are constrained from doing so because high taxation provokes intense scrutiny.
11
Patronage expenditures are determined by the product of the tax rate, t, taxable
income, Y, and the proportion of revenue which can be embezzled for patronage, e. In
turn, the rate of embezzlement is constrained by the degree of scrutiny, which is
determined by the rate of taxation.
This implies a maximum revenue available for patronage, somewhat analogous to a
Laffer curve. The maximum is determined by:
Pmax = Max e·t·Y.
(1)
wrt t
subject to e = e(t), e’<0.
To see the implication at its simplest, we linearize the inverse relationship between
the embezzlement rate and the tax rate:
e = α(1-t).
(2)
The society has an underlying rate of embezzlement, α, which is curtailed by taxation.
This underlying rate may differ greatly between societies, being determined by culture
and history. Our analysis is pertinent in those societies where α is sufficiently high
that taxation is material.
However, evidence on public corruption (Friedman,
Johnson, Kaufmann and Zoido-Lobaton, 2004) suggests that many societies do not
have strong intrinsic defenses.
12
The decision problem for the corrupt politician is thus:
max: α(1-t)·t·Y.
(3)
wrt t
At the patronage-maximizing tax rate, t* = 0.5, the resources available for patronage
are:
P = Yα/4,
(4)
Thus, in this simple model, in a society with no intrinsic defenses (α=1) a dishonest
politician would set taxes so as to generate public revenues that were half of GDP,
and half of these revenues would be embezzled for political patronage. The scrutiny
provoked by this level of taxation would defend the remaining half of the revenues
from embezzlement and the politician would, albeit reluctantly, find that the most
electorally cost-effective remaining use for these revenues was to spend them on the
provision of public goods. More generally, the provision of public goods is:
G = (2 – α)Y/4.
(5)
Competitive electoral politics drives parties to adopt the most cost-effective strategy
of winning votes, subject to the constraint imposed by endogenous scrutiny. Although
politicians would like to retain the rents for themselves, in a competitive equilibrium
parties must embezzle as much public revenue as possible and use it to bribe voters.
13
Hence, (4) and (5), though extremely simple, describe the equilibrium outcome of
electoral competition with the restraint of scrutiny endogenized.
The simplicity of the analysis enables the complication of resource rents to be
introduced without the model becoming either intractable or opaque. We denote
resource rents as a proportion of income by, r. The rents accrue directly to the
government, augmenting its revenue from the taxation of citizens. We assume that the
government is not able to ring-fence the revenue from resource rents from the
prevailing public scrutiny of its tax revenue. This is not an unreasonable assumption.
Citizen scrutiny amounts to the creation of processes such as budgets and audits
which apply to all revenues regardless of their source: the government cannot choose
which revenues are subject to scrutiny and which are not. The government is thus not
free simply to spend all the resource rents on patronage. However, unlike taxation, the
resource rents do not themselves provoke citizen scrutiny. This assumption is
precisely analogous to that made in respect of aid by Brautigam and Knack (2004)
and Moore (1998): citizens are only provoked into scrutiny by having money taken
from them by government. Government revenue thus becomes:
[t(1-r) + r]·Y,
(6)
and the maximum patronage resources available to the government become:
max α (1-t)·[t(1-r) + r]·Y
(7)
wrt t
14
The patronage-maximizing tax rate is now:
t** = (1-2r)/(2-2r).
(8)
A corollary of (8) is that the higher is revenue from resource rents the lower is the tax
rate. This relationship is manifest in developing countries: for example, the oil-rich
economies have very low non-oil taxation. While this is usually interpreted as
reflecting the reduced need for such taxation, it is notable that the same relationship
does not seem to hold in the oil-rich developed economies. For example, Norway,
which is the most oil-rich society in the OECD, has among the highest rates of non-oil
taxation. Our model provides an alternative explanation for the phenomenon: where α
is high (unlike in Norway), governments of oil-rich countries consciously set low tax
rates so as not to provoke scrutiny of the oil revenues.1 In turn, this implies that the
level of scrutiny is lower and so the rate of embezzlement is higher. More
surprisingly, total chosen revenue as a share of income, v, is constant over the entire
range 0≤r≤0.5:
v** = t**(1-r) +r = [(1-2r)/(2-2r)]·[(1-r) + r] = 0.5
(9)
That is, until taxation is driven down to zero which occurs once resource rents exceed
half of GDP. For a given total income, revenue for patronage rises as a result of
resource rents not because the government commands more money, but because it is
able to raise the same money while arousing less public scrutiny. As a result, less
1
Half of all Nigerian oil revenues accrue to the Governors of the 36 states, all of which also have
powers of local taxation: differences among states thus constitute a ‘natural experiment’. In 2006 the
Chief Economic Advisor to the President observed to Collier that the more corrupt the Governor the
lower the tax rate he chose to set.
15
needs to be diverted to the provision of public goods. A corollary of this is that over
this range of resource rents, comparing two societies with the same level of income
but with different shares of natural resource rents, the one with the higher share will
have the worse provision of public goods.
Whether a resource discovery which augments income will nevertheless worsen the
provision of public goods depends upon the scale of the resource discovery, r, and the
value of α. To see this it is useful to consider a resource discovery which precisely
doubles national income, so that once the society gets the resource rents r = 0.5. From
(8) at this point the tax rate on the non-rent economy has been driven down to zero.
Total government revenue has thus doubled: the state previously received half of
national income and now it receives all the rents, worth the entire previous national
income, but nothing else. The demise of taxation increases the rate of embezzlement
from α/2 to α. Hence, public goods provision in the presence of the rents, Gr is:
Gr = (1 – α)2Y.
(10)
Comparing this with (5), there is a critical level of α above which public goods
provision actually deteriorates, the critical rate being α = 0.857. For resource
discoveries beyond r = 0.5 there is no further scope for the reduction in taxation
(unless, for example, sinecures in public employment are introduced), and so public
goods provision unambiguously begins to improve. As noted, Norway is an example
of a relatively small resource discovery in a society with a strong prior tradition of
scrutiny, so that α was very small. Saudi Arabia is an example of a society were the
resource discovery is so large that even though α is high, the provision of public
16
goods has improved. Nigeria is an example of a society with a moderate-size
discovery and a high initial value of α, where the discovery has indeed probably
worsened public goods provision: across a range of social indicators Nigeria is ranked
below other African economies without resource rents.
While the direct focus of this model is on the provision of public goods, it has an
ready extension to a more general measure of economic performance. Public goods
are not predominantly consumed, they are in large measure capital expenditures
which cannot easily be substituted by the private sector, such as education and
infrastructure. Further, the entire model could readily be reformulated to include the
provision of public policies which do not require expenditure but do require political
effort. Politicians with dissonant interests would prefer the easy life but can be
disciplined by tax-provoked scrutiny to supply good policies. In democracies without
resource rents the costs of political effort become costs such politicians have to pay in
order to raise revenues, some of which can then be embezzled. In equivalent resourcerich democracies politicians get the same total revenue with lower taxes and so are
able to get away with lower policy effort. Thus, the result that for given income those
resource-rich democracies that lack exogenously given scrutiny mechanisms, (a low
value of α), will have inferior public goods to similar resource-scarce democracies,
readily extends to the entire range of public policies. A corollary is that ceteris
paribus they would have inferior economic growth performance. Growth performance
thus provides the most general testable formulation of our model. It predicts that, in
the absence of exogenously low values of α, resource-rich democracies would have
significantly slower growth than resource-scarce democracies. Further, since
autocracies are freed from the electoral competition that remorselessly drives such
17
democracies to an equilibrium in which public goods and policies are inferior,
resource-rich autocracies might out-perform equivalent democracies. These are the
propositions which we now test.
3. Natural Resource Rents and Democracy: Descriptive Statistics
In order to test the relationship between resource rents, democracy, and growth, it is
necessary to have a measure of resource rents. Usually, this concept has been proxied
in the literature by primary commodity exports. However, this is a highly imperfect
approximation and data are now available to enable the construction of a more
accurate measure. In this Section we describe how this measure can be built, countryby-country and year-by-year, for the period 1970-2001. We then match this data
against a quantitative measure of political rights for the same period.
Since we try to proxy ‘rents’ we did not want to rely on the commonly used SachsWarner measure of natural resources which is the ratio of primary commodity exports
to GDP. Evidently, the share of export earnings accruing as rents differs radically
both between commodities and over time depending upon the level of world prices.
For example, when world coffee prices are low coffee exports will not generate rents
whereas when oil prices are high most of the revenue will be rents. We therefore
adopted a more precise measure of rents, using environmental economic data from the
World Bank which included both costs of production and prices and so enabled us to
calculate natural resource rents as a percentage of GDPii. This calculation included
several different steps. First, we defined rents as the difference between the natural
resource price and the extraction costs. For example, for oil the World Bank database
18
provides the average of four spot crude oil prices. Prices are global, thus they vary
over time but are the same across countries. Extraction costs on the other hand vary
over time as well as across countries. In a second step, we multiplied the natural
resource rents per unit of output by the total volume extracted. We then added these
total rents for a variety of natural resources: oil, gas, coal, lignite, bauxite, copper,
iron, lead, nickel, phosphate, tin, zinc, silver and gold.iii For each year we divided the
sum of resource rents by GDP. Our regression analysis uses four year averages, so we
averaged the data over 8 sub-periods: 1970-73, 1974-77, … , 1998-2001. We were
able to construct this rent variable for 969 panel data observations. A histogram of the
natural resource rents as a percentage of GDP shows a heavily skewed frequency. A
number of countries did not extract any of these natural resources (158 observations)
and a large number only had small rents of less than one percent (363 observations).
For 180 observations the natural resource rents were between one and five percent
and 79 observations had rents between five and ten percent. We define countries with
a natural resource rent percentage of ten or higher as high rent countries. Only 187
observations were in this range.iv
We proxy democracy by the Polity IV scoring of ‘Democracy’. This is an 11 point
ordinal scale, ranging from zero to ten. Higher values indicate a greater competition
and openness of the democratic process. Although the measure is termed
‘democracy’, its criteria are essentially focused on electoral processes. Data are
available for 1,004 observations. We measure democracy at the beginning of each
sub-period. Since the democracy score is ordinal, all uses that treat it as cardinal are at
best approximations. In our subsequent regression analysis we check the robustness of
results that assume cardinality by replacing the democracy score with a binary
19
measure partitioned by a threshold. Nevertheless, descriptive statistics that assume
cardinality are a convenient introduction to the data. In Table 1 we show the means
and standard deviations for the democracy scores. The first row provides these
descriptive statistics for the entire sample period (1970-1998). The average
democracy score for the entire sample is about 4, whereas for countries with a high
percentage of natural resources it is only around 1.5. However, the standard deviation
is large compared with the entire sample, indicating that there is a wide dispersion of
democracy scores among these countries. On average democracy scores have
increased over time: for the entire sample the biggest increase occurred between 1986
and 1990 with the collapse of the Eastern Block. For the natural resource rich
economies the increases have been less marked: by the end of the period their score
was still only 1.9 as compared with 5.3 for the average country.
--- Table 1 about here ---
4. Empirical Analysis
Whether resource rents enhance or undermine the economic consequences of
democracy is a priori ambiguous. We now use our measure of rents to investigate the
issue empirically. We adopt the medium-term growth rate of the economy as our
measure of economic performance. Since resource rents are largely depletable, the
central policy issue in resource-rich societies is the transformation of depleting rents
into more sustained forms of income. We take four-year periods as our units of
20
observation to smooth out the noise of annual observations of growth rates. Evidently,
in addition to any interaction effects with the political process, natural resource rents
can be expected to have direct effects on growth and we will control for them.
We start from a simple specification which includes only the variables directly of
interest, - the level of natural resource rents, and the level of democracy, - and a single
conditioning variable, the level of per capita income (Table 2, column 1). Countries
with an initially higher democracy score have on average higher growth rates. The
coefficient on natural resource rents is insignificant. From this base we introduce the
interaction term rents·democracy which is the focus of our analysis (column 2). The
interaction term is negative and significant at the ten percent level. Democracy
appears to enhance growth except in the presence of substantial natural resources.
Around this simple specification we first investigate three variants. We allow for the
possibility of diminishing returns to rents (column 3) but find no evidence of such
non-linear effects.
Second, we allow for lagged effects. The large case-study
literature on natural resource rents has many examples of public expenditure being
increased to unsustainable levels. When we lagged resource rents as a further
explanatory variable (column 4), the term is significant, negative and substantial:
resource rents indeed appear to generate unsustainable increases in the level of output.
Third, since contemporaneous natural resource rents have no significant direct effect
in this regression we investigate dropping the term in favor of this lagged effect
(column 5). At this stage the interaction of democracy and resource rents is negative
but not significant at conventional levels (p=0.166).
--- Table 2 about here --21
While democracy is most commonly understood in terms of competitive elections
which determine how a government acquires power, a mature democracy also
includes checks and balances which in various degrees constrain how a government
can use power. This distinction is indeed central to the theory of Section 2. With an
unconstrained government, electoral competition drives an economy into patronage
politics. Only the checks and balances that are provoked by taxation force political
actors to compete through the effective provision of national public goods. The role of
taxation is important for the emergence of checks and balances because, unlike
electoral competition, these themselves are public goods that are otherwise likely to
be radically under-supplied. Elections are easy to introduce, as is evident from those
held under the precarious conditions of Afghanistan and Iraq in 2005. This is partly
because they are one-off events, and also because the incentives for parties to
participate are strong: participation is the route to power. By contrast, checks and
balances are not events but processes which must function continuously and so do not
lend themselves to short but focused efforts. Further, not only are they public goods
that nobody has a strong incentive to provide, but the actors with the power to create
them have an active interest to resist them. Precisely because checks and balances are
far more difficult to provide than electoral competition, democracies differ
considerably in the balances between these two defining features. We therefore
introduce a specific measure of checks and balances into the analysis in addition to
the more general measure of democracy. We proxy the power of checks and balances
by a measure used by Keefer and Stasavage (2004) termed ‘checks’. As implied, this
focuses on the ability of other agents to restrain the government. The index ranges
from one (few veto players) to 17 (high number of veto players). Unlike the
22
democracy index it is in principle cardinal, being a count of the number of veto
players. Although the democracy score and the checks variable are correlated
(ρ=0.72), there is sufficient variation to include the two indices in the same
regression. As can be seen from Figure 1 in the Appendix and consistent with our
hypothesis that checks are considerably more difficult to provide than elections, there
are no countries which have a low democracy score and a high number of checks.
However, there is considerable variation across countries with above average
democracy scores. Some have few or no veto players, whereas others have many: the
country with the highest number of checks (17) is India, which also has a high
democracy score.
Having distinguished between the two aspects of democracy we introduce an
additional interaction term between resource rents and checks and balances. We also
add the variable ‘checks’ itself, to control for any direct effect that it might have other
than through its effect on the utilization of resource rents, the results being shown in
Table 2, column 6. While the direct effect of checks is insignificant, the interaction of
resource rents with checks and balances is positive and significant. Further, the
adverse interaction effect of democracy and natural resources now becomes highly
significant. Thus, whereas democracy per se is distinctively detrimental for resourcerich countries, checks are distinctively beneficial. In column 7 we drop the
insignificant direct effect of checks with no change in the overall results. We now
subject these results to four tests for robustness.
First, we replace democracy as a continuous variable with a dummy which takes the
value of unity if the democracy score is greater than or equal to five (column 8). Our
23
results remain unaffected: in particular, the interaction between democracy and
natural resource rents is still negative and significant at the one percent level.
Second, we control for fixed effects. Since our model contains time invariant
variables (continent dummies) and variables which are in general only changing
slowly over time (political economy variables) we cannot estimate this model by
conventional fixed effects. However, the technique of least squares dummy variables
is equivalent to fixed effects (Hendry, Johansen and Santos., 2004). Following this
procedure, we first included one third of all country dummies in our model as
presented in column 7 and then repeated the estimation with each of the other thirds
of the country dummies. Based on these three regressions we gathered all those
country dummies which were significantly different from zero and re-estimated the
model.v These results are shown in column 9. The coefficients and standard errors are
similar to our pooled OLS regression and our core results remain significant at the one
percent level. The main difference between the pooled results and the one including
country dummies is that the coefficient on GDP per capita is slightly larger and the
standard error is smaller, making GDP significant at the five percent level.
Third, we re-introduce the contemporaneous, direct effect of resource rents (column
10). Although our baseline regression of column 7 is close to being a differences-indifferences specification, the direct effect of natural resource rents is only included
with a lag whereas the interaction term measures resource rents contemporaneously.
We therefore replace lagged by contemporaneous natural resources so that the
requirements for differences-in-differences are fully met. The coefficient on the
interaction term is unchanged and is again statistically significant at one percent.
24
Fourth, we attempt to allow for the potential endogeneity of democracy. We should
note that since we are only concerned with the interaction effect of democracy, rather
than its direct effect, this problem is less serious than were we attempting to infer a
direct causal connection from democracy onto some outcome. The interaction effect is
analogous to a difference-in-differences approach: does a difference in democracy
have a different effect depending upon resource rents? However, to address remaining
concerns we instrument for democracy. We use as an instrument the historical data on
settler mortality (Acemoglu, Johnson and Robinson, 2000). Although Acemoglu,
Johnson and Robinson use settler mortality to instrument for other types of
institutions, settlers had an interest in encouraging political institutions that were
representative of their interests, and as elsewhere, such institutions invite subsequent
pressure to expand the franchise. Indeed, in their more recent work they argue that
democracy is the outcome of a long institutional trajectory, rather than being
generated by economic development (Acemoglu et al. 2008). Because of data
limitations this instrumentation drastically reduces the size of our sample and so we
economize on other variables. Nevertheless, so instrumented, the interaction of
democracy and natural resource rents is negative and significant (Table 3).vi
--- Table 3 about here ---
We also examined the possible endogeneity of our natural resource measure. An
unobserved variable may be causing slow growth and low GDP. Since we measure
natural resource rents as a percentage of GDP, it is possible that our estimates suffer
from endogeneity bias. We tried two different approaches to tackle this issue. First,
25
we use total rather than relative resource rents. The interaction term with democracy
is negative and significant while the interaction term with checks and balances is
positive and significant. Second, we use a product of a commodity price index and the
quantity of sub-soil assets (Collier and Goderis, 2007a) as an instrument for natural
resource rents. They argue that prices and the occurrence of natural resources can be
considered as exogenous. Using this instrument the interaction terms are significant
and have the same signs as before.vii
These checks suggest that there is indeed a causal mechanism from the interaction of
natural resources and democracy onto growth as hypothesized in Section 2, and
provide some support for our choice of the model in column 7 as our baseline.
Applying the coefficients on the two critical interaction terms, in a developing
country at the 75th percentile of the democracy score, (9), but with no checks and
balances, each additional percentage point of GDP from natural resource rents reduces
growth by 0.23 percentage points. For a given level of checks and balances, resource
rents are more damaging if the country is democratic: taking a developing country
with resource rents equal to 20% of GDP, if the country is switched from being at the
75th percentile of the democracy score (9) to the 25th percentile (0), its growth rate
increases by 2.2 percentage points. By contrast, again for a given level of checks, in
the absence of resource rents democracy is good for growth: taking a developing
country without resource rents, if it is switched from the 25th percentile of the
democracy score to the 75th percentile, its growth rate increases by 1.31 percentage
points. The critical level of natural resource rents beyond which democracy becomes
dysfunctional for growth, for given checks, is 7.44 percent of GDP. While these are
the central estimates from variables that are statistically significant, we should note
26
that the confidence intervals are nevertheless substantial. For example, the reduction
of 0.23 percentage points due to an additional one percent resource rents has a 95%
confidence interval of 0.12 to 0.34 percentage points.
Within the basic structure of electoral competition being distinctively detrimental and
checks and balances being distinctively beneficial, we now investigate the routes by
which resource rents undermine the economy and hence the behavior that checks and
balances inhibits. Our approach is to control for possible routes to see whether the
interaction effects lose economic and statistical significance. The regressions with
these controls are presented in Table 4, and for ease of reference Table 4, column 1
repeats are baseline regression.
--- Table 4 about here ---
It is known that democracy tends to reduce the share in investment in GDP (Tavares
and Wacziarg, 2001) and that countries with a high dependency on natural resources
have a low investment share (Gylfason and Zoega, 2006). Since investment is likely
to be central to the transformation of resource rents into sustained growth, potentially
democracy is detrimental due to underinvestment. To test this we control for the share
of investment in GDP (Table 4, column 2). Unsurprisingly, investment is positive and
significant at the one percent level. However, its inclusion has virtually no effect. The
coefficients on both the democracy·rents interaction and the checks·rents interaction
barely change value and remain significant. Thus, to the extent that growth is driven
by investment, democracy must be undermining growth in the resource-rich countries
through the quality of investment rather than its quantity.
27
We next control for public consumption (Table 4, column 3). As is commonly found
in the growth literature, government consumption expenditure is negative. However,
this has no effect on either the significance of the core interaction terms, or the
magnitude of their coefficients. Thus, the route by which democracy undermines the
growth effects of resource rents is not that public spending becomes inflated. Again,
to the extent that public spending matters for the growth process, resource rents must
be undermining the quality of spending rather than inflating its quantity. This result is
consistent with the somewhat counter-intuitive theoretical prediction of (9) above:
resource rents induce a shift in the composition of public spending away from public
goods towards patronage goods, rather than an increase in overall spending. These
results are also consistent with the corollary proposition that resource rents are used to
reduce taxation.
We next control for the effect of ethnic diversity. Previous studies have found that
ethnic diversity is detrimental to growth (Easterly and Levine, 1997), but that this
effect is reduced by democracy (Collier, 2000). This benign interaction effect of
democracy is thus the opposite of the malign effect we have so far found. A probable
explanation for it is that autocracy is liable to be particularly damaging in the context
of ethnic diversity: if power is narrowly based on an ethnic support group
redistribution dominates the public good of growth. We now bring the two effects
together (Table 4, columns 6-9). We use a new measure of diversity proposed by
Alesina et al (2003), which classifies ethnic groups according to their racial and
linguistic characteristics. Consistent with previous research, the direct effect of ethnic
diversity on growth is adverse, and its interaction with democracy is positive.
28
However, controlling for these effects has no effect on either the significance or the
size of the coefficients of the two core interaction terms. Thus, whatever the route by
which natural resource rents undermine the growth process in a democracy, it does
not run through ethnic diversity.
Finally, we test our results against the superficially similar ‘voracity effect’ of Tornell
and Lane (1999). They also predict adverse consequences of resource rents on growth
in the context of patronage politics. However, in their model the problem is generated
by the uncoordinated ‘gauging’ of multiple powerful groups, restrained only by
concern for a participation constraint. Resource rents ease the participation constraint
and induce an increase in gauging greater than the value of the rents, this being the
voracity effect. A testable difference between the two explanations is that the voracity
effect is at its peak when there are only two powerful groups in the society, each able
to gauge. In this situation the cost of rent extraction is predicted to be double the value
of the rents themselves. There is no voracity effect either when the political system is
autocratic, or when it is fully competitive with multiple groups each holding some
power. In contrast, our model predicts that as electoral competition increases the costs
of resource rents continue to mount. The Tornell-Lane model thus predicts that the
effect of the political system on the use of resource rents is non-monotonic. We test
for this by adding the square of the democracy score interacted with resource rents as
an additional variable (Table 4, column 5). Evidently, since the democracy score is
ordinal, all rank-preserving transformations have equal validity. However, for the
Tornell-Lane hypothesis to hold this term should be significantly positive. In fact it is
completely insignificant. We have also tested using dummy variables to capture break
points in the democracy scores and found no evidence that the relationship is non-
29
monotonic. There is no sign that as political competition intensifies the problem of
resource rents is diminished, rather it continues to get worse.
We now turn to our prediction that in resource-rich democracies the mechanisms of
scrutiny would be systematically weakened: due to low taxation citizens would not be
provoked into supplying the public good of scrutiny. We test whether checks and
balances are differentially eroded by resource rents. For completeness, we also
analyze whether resource rents tend to reduce democracy itself, an effect already
established for oil economies by Ross (2001).
We begin with a simple OLS specification in which the level of checks and balances
and democracy are each explained by the level of per capita GDP, and time dummy
variables, and the lagged value of natural resource rents (Table 5). For both checks
and balances and democracy the lagged value of resource rents is highly significant
and negative. Further, as the lag is progressively lengthened from one period (four
years) to two periods (eight years) to seven periods (28 years), the significance level
and the size of the coefficient increase. The effects are large: after 28 years a country
with mean income but with resource rents worth 30% of GDP would have a checks
score in the 22th percentile instead of in the 34th percentile, and a democracy score in
the 25th percentile instead of in the 40th percentile.
--- Table 5 about here ---
30
While the OLS results are suggestive, they are open to multiple interpretations. In
Table 6 we check robustness by switching the dependent variable to the changes in
checks and democracy, respectively, over various periods, controlling for both their
initial level and per capita GDP. Again, resource rents significantly erode both checks
and democracy. These results are consistent with the predictions of the model and also
with Ross (2001).
---Table 6 about here ---
Overall, our results suggest that the form of democratic polity best-suited to resourcerich countries is one with checks and balances that are strong relative to electoral
competition. This is indeed the form of democracy in the most striking exception to
generally adverse combination of democracy and resource rents, namely Botswana.
Electoral competition is in practice quite limited: the government has never been
defeated at the polls. Yet, perhaps because the democracy has been continuous since
independence, the legal and bureaucratic procedures that constitute checks and
balances have been maintained. Other examples of democracies that have had
relatively strong checks and reasonable economic performance are Papua New
Guinea, Chile and Mexico. These polities contrast with many of the ‘instant’ and
often externally driven democracies that swept across Africa and Central Asia
following the collapse of the Soviet Union. This is indeed the form of democracy in
the most striking exception to generally adverse combination of democracy and
resource rents, namely Botswana. Electoral competition is in practice quite limited:
the government has never been defeated at the polls. Yet, perhaps because the
31
democracy has been continuous since independence, the legal and bureaucratic
procedures that constitute checks and balances have been maintained.
5. Conclusion
Resource-rich countries have tended to be autocratic and also have tended to use their
resource wealth badly. The neoconservative agenda of promoting democratization in
resource-rich countries thus offers the hopeful prospect of a better use of their
economic opportunities. Our analysis has tested whether this hopeful prognosis is
likely to be borne out.
We first showed that a priori the effect of natural resources on the economic
consequences of democracy is ambiguous. While there are plausible mechanism that
would support the proposition that resource rents enhance the benefits of democracy,
the opposite might also hold. We set out a simple model of democratic politics in
which we distinguish between two dimensions of democracy, electoral competition
and checks and balances. By undermining checks and balances, resource rents unleash
patronage politics and in these conditions electoral competition is economically
damaging.
Using new data on the value of resource rents, we then tested these propositions. We
found that in developing countries the combination of resource rents and democracy
has been significantly growth-reducing. In the absence of resource rents democracies
32
outperform autocracies, in the presence of large resource rents autocracies outperform
democracies. We found that this result was robust to controlling for the potential
endogeneity of democracy and was also robust to fixed effects. We found that the
antidote to these adverse effects of democracy was intensified checks and balances.
While countries with large resource rents need checks and balances, this is not what
they get. Resource rents tend gradually to undermine checks and balances. Thus, in
those developing societies where the state has most command over resources, the
democratic process has been least effective at controlling them for the public good.
The implication for the neoconservative agenda is that it either needs to be scaled
down or scaled up. On the criterion of economic performance targeting electoral
competition on the resource-rich societies appears to be particularly inappropriate
unless it is complemented by checks and balances. Unfortunately, whereas electoral
competition is easy to establish since there are strong incentives for participation,
checks and balances are public goods liable to be undersupplied.
33
Tables
Table 1: Democracy Scores
Period
Sample
1970-1998
1970
1974
1978
1982
1986
1990
1994
1998
4.03 (4.26)
3.29 (4.16)
3.08 (4.22)
3.18 (4.28)
3.43 (4.29)
3.72 (4.35)
4.52 (4.27)
5.29 (3.96)
5.26 (3.98)
High Natural Rents
Countries
1.46 (3.11)
0.96 (2.56)
0.89 (2.56)
1.32 (3.09)
1.76 (3.41)
1.28 (3.08)
1.89 (3.49)
2.00 (3.48)
1.92 (3.43)
Notes: Standard Deviation in parentheses.
34
Table 2: Growth, Democracy and Natural Resource Rents
ln GDP
Nat. Resources
Democracy
(1)
-0.045
(0.702)
-0.027
(0.154)
0.089
(0.036)**
NatRes·Dem
NatRes2
(2)
-0.130
(0.284)
-0.013
(0.572)
0.131
(0.005)***
-0.007
(3)
-0.118
(0.333)
0.027
(0.538)
0.141
(0.003)***
-0.010
(4)
-0.254
(0.041)**
0.053
(0.140)
0.162
(0.001)***
-0.009
(5)
-0.183
(0.146)
(6)
-0.216
(0.115)
(7)
-0.219
(0.104)
(8)
-0.137
(0.259)
(9)
-0.274
(0.043)**
0.129
(0.005)***
-0.005
0.151
(0.005)***
-0.020
0.145
(0.002)***
-0.020
0.971
(0.364)***
-0.136
0.119
(0.012)**
-0.018
(10)
-0.223
(0.083)*
-0.036
(0.280)
0.167
(0.000)***
-0.020
(0.096)*
(0.028)**
-0.001
(0.432)
(0.054)*
(0.166)
(0.003)***
(0.002)***
(0.047)***
(0.004)***
(0.002)***
-0.055
(0.016)**
-0.026
(0.067)*
-0.052
(0.044)**
-0.024
(0.805)
0.034
-0.051
(0.042)**
-0.051
(0.038)**
-0.070
(0.002)**
0.033
0.027
0.035
0.029
(0.033)**
2.901
(0.000)***
0.648
(0.292)
0.155
(0.750)
0.476
(0.333)
-1.198
(0.006)***
(0.055)*
2.918
(0.000)***
0.595
(0.333)
0.168
(0.730)
0.598
(0.216)
-1.156
(0.009)
(0.024)**
3.451
(0.000)***
0.156
(0.822)
-0.628
(0.281)
0.302
(0.556)
-1.152
(0.018)**
(0.082)*
2.994
(0.000)***
0.563
(0.330)
0.111
(0.828)
0.587
(0.228)
-1.047
(0.016)**
720
0.159
720
0.154
720
0.337
729
0.149
NatRes t-1
Checks
NatRes·Checks
East Asia
E&C Europe
MEast&NAfrica
South Asia
SSAfrica
Observations
R-squared
3.080
(0.000)***
0.688
(0.212)
0.708
(0.124)
0.683
(0.169)
-0.763
(0.063)*
2.989
(0.000)***
0.545
(0.318)
0.476
(0.326)
0.433
(0.382)
-0.921
(0.024)**
2.943
(0.000)***
0.541
(0.326)
0.462
(0.341)
0.498
(0.317)
-0.888
(0.031)**
3.055
(0.000)***
0.649
(0.271)
0.226
(0.647)
0.556
(0.240)
-1.038
(0.015)**
3.113
(0.000)***
0.719
(0.228)
0.577
(0.206)
0.700
(0.142)
-0.933
(0.030)**
(0.043)**
2.905
(0.000)***
0.645
(0.293)
0.144
(0.770)
0.499
(0.329)
-1.199
(0.006)***
858
0.138
858
0.142
858
0.145
760
0.149
760
0.142
720
0.159
Notes: Dependent variable: average annual growth. Robust p values in parentheses, * significant at 10%; ** significant at 5%; *** significant at 1%, regressions include time
dummies (not reported)
Table 3: Growth, Democracy and Natural Resource Rents – 2SLSQ
Dependent Variable
ln GDP
Nat. Res.
1st stage
Democracy
0.977
(0.000)***
0.019
(0.294)
Democracy
ln Settler Mortality
-0.560
(0.004)***
-0.011
NatRes·Democracy
Residual
East Asia
MEast&NAfrica
South Asia
SSAfrica
Observations
R-squared
2nd stage
Growth
-0.495
(0.370)
0.036
(0.161)
0.542
(0.156)
-3.963
(0.000)***
-4.903
(0.000)***
1.606
(0.025)
-1.876
(0.000)***
435
0.49
(0.012)**
-0.486
(0.207)
5.406
(0.000)***
2.908
(0.116)
0.244
(0.817)
-0.079
(0.935)
434
0.19
Notes: 2SLQ regression. Robust p values in parentheses, * significant at 10%; ** significant at 5%; ***
significant at 1%, regressions include time dummies (not reported)
Table 4: Growth, Democracy and Scrutiny
ln GDP
Nat. Resourcest-1
Democracy
NatRes·Dem
NatRes·Checks
(1)
-0.219
(0.104)
-0.051
(0.042)**
0.145
(0.002)***
-0.020
(2)
-0.511
(0.001)***
-0.052
(0.105)
0.147
(0.003)***
-0.018
(3)
-0.160
(0.269)
-0.050
(0.052)*
0.152
(0.003)***
-0.020
(4)
-0.166
(0.240)
-0.050
(0.044)**
-0.049
(0.486)
-0.020
(5)
-0.216
(0.106)
-0.051
(0.044)**
0.142
(0.003)***
-0.035
(0.002)***
0.033
(0.033)**
(0.002)***
0.032
(0.036)**
1.485
(0.000)***
(0.001)***
0.034
(0.029)**
(0.003)***
0.035
(0.036)**
(0.264)
0.033
(0.034)**
ln Investment
Gov cons.
-0.034
(0.216)
Ethnicity
-3.901
(0.000)***
0.401
(0.001)***
Ethnicity·Dem
NatRes*Dem2
East Asia
E&C Europe
MEast&NAfrica
South Asia
SSAfrica
Observations
R-squared
2.901
(0.000)***
0.648
(0.292)
0.155
(0.750)
0.476
(0.333)
-1.198
(0.006)***
720
0.159
2.173
(0.000)***
0.299
(0.633)
0.135
(0.771)
0.158
(0.747)
-0.723
(0.103)
686
0.197
2.811
(0.000)***
0.721
(0.243)
0.320
(0.520)
0.412
(0.403)
-1.053
(0.024)**
713
0.163
2.954
(0.000)***
0.528
(0.390)
-0.037
(0.939)
0.604
(0.244)
-0.244
(0.598)
718
0.181
0.002
(0.584)
2.959
(0.000)***
0.717
(0.225)
0.173
(0.721)
0.503
(0.300)
-1.175
(0.007)***
720
0.159
Notes: Dependent variable: average annual growth. Robust p values in parentheses, * significant at
10%; ** significant at 5%; *** significant at 1%, regressions include time dummies (not reported)
37
Table 5: Checks and Natural Resource Rents
ln GDP
NatRes
(1)
0.511
(0.000)***
-0.029
(0.000)***
ln GDP t-1
(2)
(3)
(4)
(5)
(6)
0.495
(0.000)***
-0.023
(0.000)***
NatRes t-1
ln GDP t-2
0.496
(0.000)***
-0.030
(0.000)***
NatRes t-2
ln GDP t-3
0.466
(0.000)***
-0.031
(0.000)***
NatRes t-3
ln GDP t-4
0.417
(0.000)***
-0.035
(0.000)***
NatRes t-4
ln GDP t-5
0.322
(0.001)***
-0.036
(0.000)***
NatRes t-5
ln GDP t-6
NatRes t-6
Observations
R-squared
(7)
758
0.306
645
0.279
518
0.259
402
0.225
294
0.172
191
0.117
0.179
(0.229)
-0.037
(0.000)***
96
0.059
Notes: Dependent variable: Checks. Robust p values in parentheses, * significant at 10%; ** significant at 5%;
*** significant at 1%, regressions include time dummies (not reported)
Table 5a: Democracy and Natural Resource Rents
ln GDP
NatRes
(1)
1.682
(0.000)***
-0.068
(0.000)***
ln GDP t-1
(2)
(3)
(4)
(5)
(6)
1.701
(0.000)***
-0.065
(0.000)***
NatRes t-1
ln GDP t-2
1.717
(0.000)***
-0.092
(0.000)***
NatRes t-2
ln GDP t-3
1.688
(0.000)***
-0.098
(0.000)***
NatRes t-3
ln GDP t-4
1.649
(0.000)***
-0.112
(0.000)***
NatRes t-4
ln GDP t-5
1.528
(0.000)***
-0.124
(0.000)***
NatRes t-5
ln GDP t-6
NatRes t-6
Observations
R-squared
(7)
762
0.487
635
0.493
506
0.518
393
0.528
287
0.524
186
0.488
1.586
(0.000)***
-0.144
(0.000)***
91
0.515
Notes: Dependent variable: Democracy. Robust p values in parentheses, * significant at 10%; ** significant at 5%;
*** significant at 1%, regressions include time dummies (not reported)
38
Table 6: Change in Checks and Natural Resource Rents
ln GDP t-1
Checkst-1
NatREs t-1
(1)
Checks–
Checks t-1
0.157
(0.003)***
-0.384
(0.000)***
-0.012
(0.000)***
(2)
ChecksChecks t-2
(3)
Checks –
Checks t-3
(4)
ChecksChecks t-4
(5)
ChecksChecks t-5
0.214
(0.004)***
-0.500
(0.000)***
-0.017
(0.000)***
ln GDP t-2
Checkst-2
NatREs t-2
0.234
(0.010)**
-0.563
(0.000)***
-0.022
(0.000)***
ln GDP t-3
Checkst-3
NatREs t-3
0.222
(0.066)*
-0.617
(0.000)***
-0.025
(0.000)***
ln GDP t-4
Checkst-4
NatREs t-4
0.120
(0.574)
-0.608
(0.018)**
-0.022
(0.014)**
ln GDP t-5
Checkst-5
NatREs t-5
ln GDP t-6
Checkst-6
NatREs t-6
Observations
R-squared
(6)
ChecksChecks t-6
626
0.182
497
0.200
381
0.190
272
0.191
168
0.197
-0.112
(0.766)
-0.415
(0.292)
-0.016
(0.352)
77
0.139
Notes: Dependent variable: Change in Checks. Robust p values in parentheses, * significant at 10%; ** significant
at 5%; *** significant at 1%, regressions include time dummies (not reported)
39
Table 6a: Change in Democracy and Natural Resource Rents
ln GDP t-1
Democracyt-1
NatRes t-1
(1)
Dem–Dem t-1
0.372
(0.000)***
-0.212
(0.000)***
-0.025
(0.000)***
(2)
Dem-Dem t-2
(3)
Dem -Dem t-3
(4)
Dem-Dem t-4
(5)
Dem-Dem t-5
(6)
Dem-Dem t-6
0.636
(0.000)***
-0.390
(0.000)***
-0.045
(0.000)***
ln GDP t-2
Democracyt-2
NatRes t-2
0.905
(0.000)***
-0.545
(0.000)***
-0.071
(0.000)***
ln GDP t-3
Democracyt-3
NatRes t-3
1.083
(0.000)***
-0.642
(0.000)***
-0.084
(0.000)***
ln GDP t-4
Democracyt-4
NatRes t-4
1.172
(0.000)***
-0.721
(0.000)***
-0.094
(0.000)***
ln GDP t-5
Democracyt-5
NatRes t-5
1.161
(0.000)***
-0.752
(0.000)***
-0.106
(0.000)***
ln GDP t-6
Democracyt-6
NatRes t-6
ln GDP t-7
Democracyt-7
NatRes t-7
Observations
R-squared
(7)
Dem-Dem t-7
710
0.149
579
0.243
472
0.349
368
0.420
268
0.463
176
0.478
1.238
(0.000)***
-0.745
(0.000)***
-0.124
(0.000)***
86
0.488
Notes: Dependent variable: Change in Democracy. Robust p values in parentheses, * significant at 10%; **
significant at 5%; *** significant at 1%, regressions include time dummies (not reported)
40
References
Auty, R. (ed.) 2001. Resource Abundance and Economic Development. Oxford,
Oxford University Press.
Acemoglu, D., S. Johnson and J. Robinson. 2000. The Colonial Origins of
Comparative Development: An Empirical Investigation, American Economic Review
91: 1369-1401.
Acemoglu, D., S. Johnson, J.A. Robinson and P. Yared. 2008. Income and
Democracy. American Economic Review, forthcoming.
Alesina, Alberto, Arnaud Devleeschauwer, William Easterly, Sergio Kurlat, and
Romain Wacziarg. 2003. Fractionalization. Journal of Economic Growth 8 (2): 155194.
Alesina, Alberto, and Eliana La Ferrera. 2004. Ethnic Diversity and Economic
Performance. Journal of Economic Literature 43:762-800.
Barro, R.J. 1996. Democracy and growth. Journal of Economic Growth 1: 1-27.
Beck, T., Clarke, G., Groff, A., Keefer, P. and P. Walsh. 2001. New tools in
comparative political economy: The Database of Political Institutions. World Bank
Economic Review 15(1): 165-176.
Besley, T., 2006, Principled Agents? Oxford University Press, New York.
Brautigam, D. and S. Knack. 2004 Foreign Aid, Institutions and Governance in SubSaharan Africa, Economic Development and Cultural Change 52: 255-285.
Collier, Paul. 2000. Ethnicity, Politics and Economic Performance. Economics and
Politics 12 (3):225-245.
Collier, Paul and Benedikt Goderis. 2007. Prospects for Commodity Exporters:
Hunky Dory or Humpty Dumpty? World Economics 8:1-15.
Collier, Paul and Benedikt Goderis. 2007a. Commodity Prices, Growth, and the
Natural Resource Curse: Reconciling a Conundrum. University of Oxford. Mimeo.
Corden, W. Max, and J. Peter Neary, “Booming Sector and De-Industrialization in a
Small Open Economy,” The Economic Journal 92 (1982), 825-848.
Drazen, A. 2000. Political Economy in Macroeconomics. Princeton, New Jersey:
Princeton University Press.
Easterly, William and Ross Levine. 1997. Africa's Growth Tragedy: Politics and
Ethnic Divisions. The Quarterly Journal of Economics 112:1203-50.
Feng, Y. 2003. Democracy, Governance and Economic Performance: Theory and
Evidence, MIT Press.
41
Friedman, E., Johnson, S., Kaufmann, D. and P. Zoido-Lobaton, 2004. Dodging the
grabbing hand: the determinants of unofficial activity in 69 countries. Journal of
Public Economics 76 (2000) 459–493.
Githongo, J. 2006. Inequality, Ethnicity and the Fight against Corruption in Africa: A
Kenyan Perspective. Forthcoming in Economic Affairs.
Gylfason, T. 2001. Natural resources, education,and economic development.
European Economic Review 45: 847-859.
Gylfason, T. and G. Zoega. 2006. Natural Resources and Economic Growth: The Role
of Investment. The World Economy 29: 1091-1115.
Hamilton, K. and M. Clemens. 1998. Genuine Saving Rates in Developing Countries,
Worldbank, Washington D.C. available on
http://lnweb18.worldbank.org/ESSD/envext.nsf/44ByDocName/GreenAccountingAdj
ustedNetSavings
Hendry, D.F., S. Johansen and C. Santos. 2004. Selecting a Regression Saturated by
Indicators, unpublished paper, Economics Department, University of Oxford.
Heston, A., Summers, R. and B. Aten. 2002. PWT 6.1. Center for International
Comparisons at the University of Pennsylvania (CICUP), October 2002.
http://pwt.econ.upenn.edu/
Jaggers, Keith, and Ted Robert Gurr. 1995. Tracking Democracy's Third Wave with
the Polity III Data. Journal of Peace Research 32:469-482.
Keefer, P. and D. Stasavage. 2003. The Limits of Delegation: Veto Players, Central
Bank Independence and the Credibility of Monetary Policy. American Political
Science Review (August).
Lipset, S. M. 1959. Some Social Requisites of Democracy: Economic Development
and Political Legitimacy. American Political Science Review 53: 69-105.
Mehlum, Halvor, Karl Moene, and Ragnar Torvik. 2006. Institutions and the
Resource Curse. Economic Journal 116:1-20.
Moore, Mick, 1998, ‘Death without Taxes: Democracy, State Capacity and Aid
Dependence in the Fourth World, in Towards a Democratic Developmental State, eds.
Gordon White and Mark Robinson, Oxford, Oxford University Press.
Olson, M. 1993. Dictatorship, Democracy, and Development. American Political
Science Review 87: 567-576.
Olsson, O. 2007. Conflict Diamonds. Journal of Development Economics 82: 267286.
Przeworski, A. and F. Limongi. 1993. Political Regimes and Economic Growth. The
Journal of Economic Perspectives 7: 51-69.
42
Robinson, James A, Ragnar Torvik, and Thierry Verdier. 2006. Political Foundations
of the Resource Curse. Journal of Development Economics 79: 447-468.
Ross, M. L. 2001. Does Oil Hinder Democracy? World Politics 53 (April):325-61.
Ross, M. L. 2004. Does Tax lead to Representation? British Journal of Political
Science 34: 229-249.
Ross, M. L. 2006. A Closer Look at Oil, Diamonds, and Civil War. Annual Review of
Political Science 9: 263-300.
Sachs, Jeffrey D., and Andrew M. Warner. 2005. "Natural Resource Abundance and
Economic Growth." In Leading Issues in Economic Development, ed. G. M. Meier
and J. E. Rauch. 8th edition. New York: Oxford University Press.
Smith, B. 2004. Oil Wealth and Regime Survival in the Developing World, 19601999, American Journal of Political Science 48 (2): 232-246.
Selden, Z. 2004. Neoconservatives and the American Mainstream. Hoover Institution
Policy Review, April 2004.
Söderbom, M., Teal, F., Wambugu, A. and G. Kahyarara, 2006. The Dynamics of
Returns to Education in Kenyan and Tanzanian Manufacturing, Oxford Bulletin of
Economics and Statistics 68: 261-288.
Tavares, J. and R. Wacziarg, 2001, How Democracy Affects Growth, European
Economic Review 45: 1341-1378.
Tilly, Charles, 1975, The Formation of National States in Western Europe, Princeton
University Press.
Tornell, Aaron and Philip Lane, 1999. The Voracity Effect, American Economic
Review 89: 22-46.
Vicente, P. 2007. Does Oil Corrupt? Evidence from a Natural Experiment in West
Africa, Centre for the Study of African Economies.
http://users.ox.ac.uk/~econ0192/oil.pdf
Wooldridge, J. M. 2002. Econometric Analysis of Cross Section and Panel Data. MIT
Press: Cambridge MA.
43
Appendix
Descriptive Statistics for the Core Model
Table A1: Means
Variable
Growth
ln GDP
Nat. Resources
Nat. Res. t-1
Democracy
Checks
Ethnic
ln Investment
Gov. Cons.
Obs
720
720
720
720
720
720
718
686
713
Mean
1.29
7.48
5.99
6.30
4.51
2.54
0.46
2.57
15.66
Std. Dev.
3.64
1.62
11.84
12.92
4.28
1.77
0.26
0.62
6.81
Min
-16.23
4.57
0
0
0
1
0.012
0.60
3.67
Max
20.29
10.73
80.60
106.60
10
17
0.93
4.03
74.32
Table A2: Correlation Coefficients
ln GDP
Nat. Res.
Nat. Res. t-1
Democracy
Checks
Ethnic
ln Investment
Gov. Cons.
growth
ln GDP
0.115
-0.101
-0.142
0.160
0.161
-0.236
0.278
-0.081
1
-0.068
-0.077
0.683
0.505
-0.512
0.687
0.354
Nat.
Res.
1
0.819
-0.219
-0.145
0.172
-0.029
0.065
Nat.Res. t-1
Democr.
Checks
Ethnic
ln Inv.
1
-0.202
-0.138
0.131
-0.056
0.064
1
0.716
-0.427
0.448
0.232
1
-0.326
0.358
0.181
1
-0.469
0.130
1
0.192
Table A3: Checks Scores
Period
Sample
1975-1998
1974
1978
1982
1986
1990
1994
1998
2.34 (1.72)
1.74 (1.41)
1.89 (1.35)
2.08 (1.51)
2.17 (1.59)
2.41 (1.89)
2.97 (1.87)
2.88 (1.87)
High Natural Rents
Countries
1.54 (1.34)
1.03 (0.16)
1.26 (0.70)
1.46 (1.10)
1.48 (1.43)
1.64 (1.89)
2.40 (2.00)
2.04 (1.59)
Notes: Standard Deviation in parentheses.
44
Sample:
Albania, Algeria, Angola, Argentina, Armenia, Austrialia, Austria, Azerbaijan,
Bangladesh, Belarus, Belgium, Benin, Bolivia, Botswana, Brazil, Bulgaria, Burkina
Faso, Burundi, Cameroon, Canada, Central African Republic, Chad, Chile, China,
Colombia, Dem. Rep. Congo, Rep. Congo, Costa Rica, Cote d'Ivoire, Croatia, Czech
Republic, Denmark, Dominican Republic, Ecuador, Egypt, El Salvador, Eritrea,
Estonia, Ethiopia, Fiji, Finland, France, Gabon, The Gambia, Georgia, Germany,
Ghana, Greece, Guatemala, Guinea, Guinea-Bissau, Haiti, Honduras, Hungary, India,
Indonesia, Iran, Ireland, Israel, Italy, Jamaica, Japan, Jordan, Kazakhstan, Kenya,
Kuwait, Kirguiz Rep., Laos, Latvia, Lesotho, Macedonia, Madagascar, Malawi,
Malaysia, Mali, Mauritania, Mauritius, Mexico, Moldova, Mongolia, Morocco,
Mozambique, Namibia, Nepal, Netherlands, New Zealand, Nicaragua, Niger,
Nigeria, Norway, Oman, Pakistan, Panama, Papua New Guinea, Paraguay, Peru,
Philippines, Poland, Portugal, Romania, Russian Federation, Rwanda, Saudi Arabia,
Senegal, Sierra Leone, Singapore, Slovak Republic, Slovenia, South Africa, Sri
Lanka, Sudan, Sweden, Syrian Arab Republic, Tanzania, Thailand, Togo, Trinidad
and Tobago, Tunisia, Turkey, Turkmenistan, Uganda, Ukraine, United Kingdom,
United States, Uruguay, Uzbekistan, Venezuela, Vietnam, Yemen, Zambia,
Zimbabwe.
45
Data Description and Sources:
Economic Growth
We used WDI 2003 data for GDP and population. GDP is measured in constant 1995
US dollars, we divided GDP by the population to calculate per capita GDP. We
approximated the growth of per capita GDP by taking the log differences at the
beginning and end of each sub period (1970-73, 1974-77, …, 1998-2001) and divided
this difference by the number of years, four, and multiplied this by 100.
GDP per capita
We measure GDP per capita at the beginning of each sub-period (1970-73, 1974-77,
…, 1998-2001). Data are measured in constant 1995 US dollars and the data source is
WDI 2003.
Natural Resource Rents
Using data from the World Bank’s adjusted savings project we calculated the rents for
each commodity by subtracting the cost from the commodity price. We then
multiplied the rents per unit by the amount extracted and summed across the different
commodities. We then calculated the share of rents in GDP. Since the rents are
provided in current US dollars we used the WDI 2003 GDP in current dollars to
calculate this share. Natural resources for which rent data were available are: oil, gas,
coal, lignite, bauxite, copper, iron, lead, nickel, phosphate, tin, zinc, silver and gold.
The data are described in Hamilton and Clemens (1998) and available from
http://lnweb18.worldbank.org/ESSD/envext.nsf/44ByDocName/GreenAccountingAdj
ustedNetSavings
Democracy
The degree of competition for executive power is measured on a scale of zero (low) to
ten (high). We used the Polity IV score at the beginning of each sub-period (1970-73,
1974-77, …, 1998-2001). Source: http://www.cidcm.umd.edu/polity/index.html. The
data are described in Jaggers and Gurr (1995).
Investment
We used total investment as a percentage of GDP and averaged and logged over each
sub-period (1970-73, 1974-77, …, 1998-2001). Data source: PWT as described in
Heston, Summers and Aten (2002).
Checks
This variable captures the number of veto players. This variable is built from several
other variables, two of which are the legislative and the executive indices of electoral
competitiveness. The checks and balances index ranges from 1 to 17 with higher
numbers indicating a higher number of veto players. Data Source: DPI2000, data are
described in Beck et al (2001) and Keefer and Stasavage (2003) and are available
from http://econ.worldbank.org/view.php?type=18&id=25467
Ethnic Diversity
Diversity is a measure of ethnic fractionalization is measured as the probability of two
randomly people not belonging to the same ethnic group. This measure of ethnic
fragmentation is based on a broad classification of groups, taking into account not
46
only language but also other cleavages such as racial characteristics. Data source:
Alesina et al (2003).
Regional Dummies
The regional dummies were obtained from Collier and Dollar (2001).
Sub-Saharan Africa: Angola, Burundi, Benin, Burkina Faso Central African
Republic Cote Congo, Rep Cape Verde Djibouti, Eritrea, Ethiopia, Gabon, Ghana,
Guinea, The Gambia, Guinea-Bissau, Equatorial Guinea, Kenya, Liberia, Lesotho,
Madagascar, Mali, Mozambique, Mauritania, Mauritius, Malawi, Namibia, Niger,
Nigeria, Rwanda, Sudan, Senegal, Sierra Leone, Somalia, Sao Tome and Principe,
Swaziland, Seychelles, Chad, Togo, Tanzania, Uganda, South Africa, Congo, Dem.
Rep., Zambia and Zimbabwe.
South Asia: Afghanistan, Bangladesh, Bhutan, India, Sri Lanka, Maldives, Nepal and
Pakistan.
East Asia: China, Indonesia, Japan, Korea, Rep., Malaysia, Philippines, Singapore
and Thailand.
Middle East and North Africa: United Arab Emirates, Bahrain, Algeria, Egypt,
Arab Rep., Greece, Iran, Islamic Rep., Iraq, Israel, Jordan, Kuwait, Lebanon, Libya,
Morocco, Malta, Oman, Portugal, Qatar, Saudi Arabia, Syrian Arab Republic, Tunisia
and Yemen, Rep.
Eastern and Central Europe: Albania, Armenia, Azerbaijan, Bulgaria, Bosnia and
Herzegovina,
Belarus,
Czech
Republic,
Estonia,
Georgia,
Croatia,
Hungary,Kazakhstan, Kyrgyz Republic, Lithuania, Latvia, Moldova, Macedonia,
FYR, Poland, Romania, Russian Federation, Slovak Republic, Slovenia, Tajikistan,
Turkmenistan, Turkey, Ukraine, Uzbekistan and Yugoslavia (Serbia/Montenegro).
47
i
During the resource boom of the 1970s, the average resource-rich country scored only 0.96 on the
Polity IV scale of political rights (the scale ranges 0-10). By the mid-1990s the score had risen to 3.47.
The scale ranges from 0-10. Details of these figures are given in Section 3.
ii
This data set has now been used by a number of studies, for example Ross (2006) analyzes the link
between the resource rents from oil and civil war.
iii
A notable omission from our coverage is diamonds. As Olsson (2007) shows diamonds have adverse
effects on economic growth conditional upon poor institutions, a result entirely consistent with our
subsequent results.
iv
For two observations this average is larger than 100. This is possible because the numerator and
denominator are based on different measurement concepts, rents and value added. In any case in our
regression analysis we could not use these observations because other data were missing.
v
We would like to thank David Hendry for this suggestion. Country dummies included are: Armenia,
Australia, Azerbaijan, Botswana, Central African Republic, Chile, China, Dominican Republic, Egypt,
Estonia, Georgia, Guinea, Haiti, Ireland, Jamaica, Japan, Lesotho, Lithuania, Latvia, Madagascar,
Macedonia, Mauritius, Niger, Nicaragua, Norway, Oman, Peru, Philippines, Portugal, Sierra Leone,
Tunisia, Uzbekistan, Vietnam, Democratic Republic of the Congo and Zambia.
vi
Due to the democracy interaction term our model is non-linear in the variables and we use the test for
endogeneity as described in Wooldridge (2002). We calculate the residuals from the first stage
regression and include them in the second stage regression. A zero coefficient on the residual is
interpreted as an indication that endogeneity does not pose major problems and that OLS regression
can be used. This approach is sometimes referred to as the ‘control function’ approach. A recent
application can be found in Söderbom et al (2006, section V).
vii
Results available upon request.
48